periodica polytechnica

Mechanical Engineering
51/2 (2007) 7175
doi: 10.3311/pp.me.2007-2.05
web: http:// www.pp.bme.hu/ me
c Periodica Polytechnica 2007
RESEARCH ARTICLE
Integration method for constitutive
equation of von Mises elastoplasticity
with linear hardening
Attila Kossa
Received 2008-01-18
Abstract
This paper presents a summary of the new semi-analytical
integration method presented in [10] for von Mises elastoplas-
ticity model with combined linear isotropic-kinematic hardening
within a small deformation range. Solutions for the case of con-
stant strain rate and constant stress rate assumptions are also
presented. Furthemore, it is shown how the general solution
reduces to the particular cases of purely kinematic hardening,
purely isotropic hardening and perfect plasticity, respectively.
Keywords
elastoplasticity combined hardening exact integration
Acknowledgement
This research has been supported by the National Devel-
opment and Research Foundation, Hungary (under Contract:
OTKA, T046488). This support is gratefully acknowledged.
Attila Kossa
Department of Applied Mechanics, BME, H-1111 Budapest, M uegyetem rkp.
5., Hungary
e-mail: kossa@mm.bme.hu
1 Introduction
The most widely used plasticity model is the von Mises
elastoplasticity. In terms of the hardening rule we can cate-
gorize the various cases as: perfect plasticity, kinematic hard-
ening, isotropic hardening and combined hardening. Analyti-
cal solution for perfect plasticity was presented in [14], [11],
[9]. Exact solution for linear kinematic hardening is derived in
[3, 11, 21, 23, 24]. A semi-analytical solution is given in [20] for
purely isotropic hardening. Due the complexity of the governing
constitutive relations corresponding to the combined hardening
case, fully analytical solution is not given in the literature. The
problem was discussed in [5], [16], [17], [24] deriving some ef-
cient approximate and nearly exact solutions. Ecient numer-
ical integration techniques can be found in [24, 6, 8, 12, 15, 16].
This paper focuses on the new semi-analytical solution of com-
bined hardening materials presented in [10] and includes some
remarks corresponding to the general solution.
The paper is organized as follows. The notation for equations
will be introduced at the end of this section. Section 2 contains
a brief review of the constitutive relations for von Mises elasto-
plasticity with combined linear isotropic-kinematic hardening.
In Section 3 the solution corresponding to constant strain rate
assumption is presented. In Section 4 the formulas will be de-
rived when the loading is dened by constant stress rate.
Regarding notation, tensors are denoted by bold-face char-
acters, the order of which is indicated in the text. The tensor
product is denoted by , and the following symbolic operations
apply: a : b = a
i j
b
i j
, and (C : a)
i j
= C
i j kl
a
kl
, with the sum-
mation over repeated indices. The superscripts T and 1 denote
transpose and inverse, respectively, and the prex tr refers to the
trace. The symbol a =

a : a is used to denote a norm of
second order tensor a. Furthermore, standard tensors are de-
noted by for the second-order unit tensors, and by I for the
symmetric fourth-order unit tensor.
Integration method for constitutive equation of von Mises elastoplasticity with linear hardening 71 2007 51 2
2 Rate-form constitutive equations of the von Mises
elastoplasticity with combined linear kinematic and
isotropic hardening
The well-known constitutive relations for the von Mises
elastoplasticity model with combined linear isotropic-kinematic
hardening are summarized here (for more details see [13], [18]).
The classical additive decomposition of the total strain-rate is
=
e
+
p
, (1)
where
e
is the elastic and
p
is the plastic strain rate, re-
spectively. The elastic behaviour is governed by the following
constitutive relation:

= D
e
:
e
, (2)
where the fourth-order elasticity tensor in linear isotropic elas-
ticity takes the form:
D
e
= 2GT + K . (3)
T = I
1
3
is the fouth-order deviatoric operator tensor, and
G and K are the shear and bulk moduli, respectively. The von
Mises yield function for combined isotropic kinematic harden-
ing is dened by
F(, , ,R) = R ( ) 0, (4)
where R ( ) represents the isotropic hardening law in terms of
a scalar plastic state variable . The so-called reduced stress
deviator is dened as
=sss , (5)
where sss =
1
3
tr is the deviatoric stress, is the back stress
describing the translation of the yield surface in the deviatoric
stress space due to the kinematic hardening. For associative ow
rule, the plastic strain rate tensor is found from the following
expression:

p
=

, (6)
where the plastic loading parameter is given by
=
_
t
0

dt. (7)
The linear isotropic and kinematic hardening moduli can be ex-
pressed in the following forms:
h
i so
= Mh, h
ki n
= (1 M) h, (8)
where h = 2H/3 and H is the constant plastic hardening mod-
ulus. The mixed hardening parameter M [0, 1] denes the
relation between the isotropic and kinematic part, respectively.
M = 0 stands for purely kinematic hardening, M = 1 for the
purely isotropic hardening. The linear isotropic hardening func-
tion is written as
R = R
0
+ h
i so
, (9)
where R
0
is a material constant related to the initial value of
yield stress R
0
=

2/3
y
. The evolution law for the back
stress tensor is dened by the Ziegler-Pragers model as

= h
ki n

p
= h
ki n

. (10)
The loading/unloading conditions can be expressed in Kuhn-
Tucker form as

0, F(, , ,R) 0,

F(, , ,R) = 0. (11)

The plastic multiplier

can be calculated using the plastic con-
sitency condition

F = 0 and the Eqs. (1) - (9):

=
2G :
R(2G +h)
. (12)
Finally the elastoplastic constitutive relations can be expressed
as

= D
ep
: , , , D
ep
= D
e

4G
2
R
2
(2G +h)
, (13)
where D
ep
is the so-called elastoplastic, or continuum tangent
modulus tensor. The constitutive equation of elastoplasticity de-
ned above can be separated into a deviatoric and a hydrostatic
part as follows:
s s s = 2G e e e
4G
2
R
2
(2G +h)
( : e e e) , tr

= 3Ktr , (14)
where e e e =
1
3
tr is the deviatoric strain rate. The rate of
the is obtained from (10), (8) and (12) as

=
2G (1 M) h
R
2
(2G +h)
( : e e e) , (15)
and the evolution law for the radius of the yield surface, com-
bining (7), (8), (9) and (12) is given by

R =
2GMh
R(2G +h)
( : e e e) . (16)
From (14) and (15), the expression for

can be written as

= 2G e e e
2G
R
2
_
1
Mh
2G +h
_
( : e e e) . (17)
3 Time integration of constitutive equations with con-
stant strain rate assumption
Here we restrict our analysis to purely elastoplastic loading,
i.e. when both the initial and the nal state lie on the yield sur-
face. The main goal is to determine the solution of Eqs. (14)-
(16) when the loading is given by constant strain rate. It is pos-
sible to dene the following inner product between the strain
rate tensor and the relative stress on the deviatoric plane (this
technique was rst proposed in [11] for perfect plasticity):
: : e = e = e = e e e cos = R e e e cos . (18)
Per. Pol. Mech. Eng. 72 Attila Kossa
2 3
1
s
n
s( ) t
x
x
a
a
e
y
n
y( ) t
R
n
R t ( )
e
n
n ( ) t
( ) t
Fig. 1. Geometric interpretation of the angle
Schematic illustration of is shown in Fig. 1. Using this
angle variable after some straightforward algebraic manipu-
lation (detailed discussion is given in [10]) we can obtain the
following expression for the radius of the yield surface in terms
of the angle :
R () = R
n
_
sin
n
sin
_
a
, (19)
where
n
is the starting value of (t ) computed at the known
n-th state (t = 0). The dimensionless parameter a is
a =
Mc
1 +c
, c =
h
2G
. (20)
Combining Eqs. (14)-(19) nally we arrive at the following for-
mula which implicitly denes the function (t ):
4G e e e t
R
n
(sin
n
)
a
= B
_
cos
2
,
1
2
,
a
2
_
B
_
cos
2

n
,
1
2
,
a
2
_
,
(21)
where the incomplete Beta function (see [1], [19]) is dened by
B (x, , ) =
_
x
0
s
1
(1 s)
1
ds, 0 x < 1.
(22)
Ecient technique for computing the inverse incomplete func-
tion can be found in [7]. After (t ) is obtained then using
trigonometric identities the (t ) solution can be expressed as
a linear combination of the relative stress
n
at n-th state (t = 0)
and the strain rate tensor e e e as
(t ) = A

n
+ B

e e e, (23)
with the constant parameters A

and B

:
A

=
_
sin
n
sin
_
a1
, B

=
R
n
sin (
n
)
e e e sin
n
_
sin
n
sin
_
a
.
(24)
Using (18), (19), (21) and (23) in (14) the solution for sss (t ) can
be obtained in the following form:
sss (t ) = A
s

n
+ B
s
e e e, (25)
where the constant A
s
and B
s
are
A
s
=
1
(a 1) (c +1)
_
1 A

_
, (26)
B
s
=
2Gc
(1 +c)
t
R
n
cos
n
e e e
A
s
+
R
n
sin
a

n
2 (1 +c) e e e
(27)
_
B
_
cos
2
,
1
2
, 1
a
2
_
B
_
cos
2

n
,
1
2
, 1
a
2
__
.
3.1 Proportional loading
The solution derived above is not applicable for proportional
loading where
n
= 0. When the loading is proportional then
the tensor e e e and are coaxial, therefore we can write:
e e e
e e e
=

R
= R
e e e
e e e
, : e : e : e = R e e e .
(28)
Substituting (28)
2
and (28)
3
in (14)
1
we obtain the solution for
the deviatoric stress:
s s s =
h
1 +c
e e e sss (t ) =sss
n
+
h
1 +c
e e et. (29)
Combining (28)
3
and (16) the solution for the radius of the yield
surface takes the form:

R =
Mh
(1 +c)
e e e R (t ) = R
n
+
Mh
(1 +c)
e e e t.
(30)
Finally the solution for the relative stress comes from (28)
2
:

t
=
n
+
Mh
(1 +c)
e e et. (31)
3.2 Case of purely isotropic hardening
Now the case where in Eq. (8) M = 1, which corresponds to
ow theory with purely isotropic behaviour of strain hardening
is considered. In this case the centre of the yield surface is still
xed, therefore there is no back-stress , i.e. sss. The angle
is dened through s : e s : e s : e = R e e e cos . The solution derived for
combined hardening reduces to the following form:
sss (t ) = A
i so
s
sss
n
+ B
i so
s
e e e, (32)
A
i so
s
=
_
sin
n
sin
_
a1
, (33)
B
i so
s
=
R
n
sin (
n
)
e e e sin
n
_
sin
n
sin
_
a
. (34)
This solution with detailed calculations is presented in [20].
3.3 Case of purely kinematic hardening
Next the ow theory with linear kinematic hardening is dis-
cussed, in which particular case the radius of the yield surface
remains unchanged and hardening occurs due only to the change
of back-stress tensor. The mixed hardening parameter in this
case is M = 0. The solution (21) of (t ) reduces to
(t ) = 2 arctan
_
_
_e
_

2G e e e t
R
0
_
tan

n
2
_
_
_. (35)
Integration method for constitutive equation of von Mises elastoplasticity with linear hardening 73 2007 51 2
According to (23) the relative stress (t ) is calculated by
(t ) = A
ki n

n
+ B
ki n

e e e, (36)
A
ki n

=
sin
sin
n
, (37)
B
ki n

=
R
n
sin (
n
)
e e e sin
n
, (38)
where the parameters A
ki n

and B
ki n

come from (26) and (27)

substituting M = 0. The solution of the deviatoric stress sss (t ) is
given by
sss (t ) =sss
n
+ A
ki n
s

n
+ B
ki n
s
e e e, (39)
A
ki n
s
=
1
(c +1)
_
A
ki n

1
_
, (40)
B
ki n
s
=
R
n
(c +1) e e e
_
cos cos
n
A
ki n

_
+
2Gc
c +1
t. (41)
3.4 Case of perfect plasticity
Without hardening (h = 0), the solution (26)-(27) reduces to
the well-known form of perfect plasticity (see [11]). (t ) can
be computed by (35) with the remark that in this case the angle
is dened between the deviatoric stress and the strain rate tensor.
The analytical solution is represented in the form
sss (t ) = A
p
s
sss
n
+ B
p
s
e e e, (42)
A
p
s
=
sin
sin
n
, (43)
B
p
s
=
R
n
sin (
n
)
e e e sin
n
. (44)
4 Time integration of constitutive equations with con-
stant stress rate assumption
In the previous section the case when the loading is given
by constant strain rate was discussed. We have the so-called
inverse problem when the loading path is prescribed in the stress
space. Now consider the case when the stress rate assumed to
be constant. The inverse of (14)
1
is written as
e e e =
1
2G
s s s +
1
R
2
h
( : s s s) . (45)
Substituting in (17) we have

= s s s
(1 M)
R
2
( : s : s : s) . (46)
Similarly to (18) here is also possible to dene the following
inner product introducing the angle variable :
: s = s = s = R s s s cos . (47)
Substituting (47) in (45) and using the consistency condition -
nally we arrive at the following solution for the radius of the
yield surface (details can be found in [10]):
R () = R
n
_
sin
n
sin
_
M
. (48)
After quite long, otherwise straightforward manipulation the
(t ) function can be implicitly dened by the following non-
linear equation:
2 s s s t
R
n
(sin
n
)
M
= B
_
cos
2
,
1
2
,
M
2
_
B
_
cos
2

n
,
1
2
,
M
2
_
.
(49)
Using trigonometric identities the relative stress (t ) can be
written as a linear combination of
n
and the stress rate tensor s s s
(t ) = A

n
+ B

s s s, (50)
A

=
_
sin
n
sin
_
M1
, (51)
B

=
R
n
sin (
n
)
s s s sin
n
_
sin
n
sin
_
M
. (52)
Note that the parameters A

and B

are dierent from thoose in-

troduced in the previous section. The solution for the deviatoric
strain takes the form:
eee (t ) =eee
n
+ A
e

n
+ B
e
s s s, (53)
A
e
=
1
2Gc (M 1)
_
A

1
_
, (54)
B
e
=
(1 +c)
2Gc
t
R
n
cos
n
s s s
A
e

R
n
sin
M

n
4Gc s s s
(55)
_
B
_
cos
2
,
1
2
, 1
M
2
_
B
_
cos
2

n
,
1
2
, 1
M
2
__
.
4.1 Proportional loading
In case of proportional loading the stress rate tensor and the
relative stress tensor are coaxial, i.e. we can write:
s s s
s s s
=

R
= R
s s s
s s s
, : s : s : s = R s s s .
(56)
Substituting (56)
2
and (56)
3
in (46) gives the solution for (t )
as

= M s s s (t ) =
n
+ M s s st. (57)
The solution for R (t ) can be obtained by combining (56)
2
, (56)
3
and (16)

R = M s s s R (t ) = R
n
+ M s s s t. (58)
After both (t ) and R (t ) are derived the solution of eee (t ) can be
found by substituting Eqs. (56)-(58) in (45)
e e e =
1 +c
2Gc
s s s eee (t ) =eee
n
+
1 +c
2Gc
s s st. (59)
4.2 Case of isotropic hardening
Solution corresponding to purely isotropic hardening is de-
rived here. In this particular case hardening occurs only through
the variation of the radius of the yield surface. Since sss ,
the angle (t ) is dened between the deviatoric stress sss and the
deviatoric stress rate s s s and it can be calculated from
(t ) = arctan
_
R
0
sin
n
s s s t + R
0
cos
n
_
, (60)
Per. Pol. Mech. Eng. 74 Attila Kossa
which formula comes from (49) by substituting M = 1. The
solutions (53)-(55) reduce to the following expression
eee (t ) =eee
n
+ A
i so
e
sss
n
+ B
i so
e
s s s, (61)
A
i so
e
=
1
2Gc
ln
_
sin
n
sin
_
, (62)
B
i so
e
=
(1 +c) t
2Gc

R
n
cos
n
s s s
A
i so
e

R
n
sin
n
2Gc s s s
(
n
) .
(63)
This solution can be found in [20].
4.3 Case of kinematic hardening
In purely kineamtic hardening case the mixed hardening pa-
rameter is M = 0. Equation (49) is simplied and the relation
for (t ) is given by
(t ) = 2 arctan
_
_
_e

s s s t
2R
0
tan
_

n
2
_
_
_
_.
The solution for the relative stress (t ) according to (50)-(52)
becomes
(t ) = A
ki n

n
+ B
ki n

s s s, (64)
A
ki n

=
sin
sin
n
, (65)
B
ki n

=
R
n
sin (
n
)
s s s sin
n
. (66)
Substituting M = 0 in formulas (53)-(55) gives the analytical
solution for the deviatoric strain as
eee (t ) =eee
n
+ A
ki n
e

n
+ B
ki n
e
s s s, (67)
A
ki n
e
=
1 A
ki n

2Gc
, (68)
B
ki n
e
=
(1 +c)
2Gc
t
R
n
cos
n
s s s
A
ki n
e

R
n
2Gc s s s
(cos cos
n
) . (69)
5 Conclusion
A brief summary of the new semi-analytical solution for von
Mises plasticity with combined linear kinematic and isotropic
hardening is given in this paper. The case of constant strain
rate and constant stress rate loading are discussed and the solu-
tions for each cases are derived. Furthermore it has been shown
how reduces all the general solution for the following particular
cases: purely kinematic hardening, purely isotropic hardening,
perfect plasticity.
References
1 Abramowitz M., Stegun I.A., Handbook of Mathematical Functions. Ap-
plied Mathematics Series, Vol. 55, Dover Publications, New York, 1968.
2 Artioli E., Auricchio F., Beiroda Veiga L., Generalized midpoint integra-
tion algorithms for J
2
plasticity with linear hardening, International Journal
for Numerical Methods in Engineering 72 (2007), 422-463.
3 Auricchio F., Beiroda Veiga L., On a new integration scheme for von-
Mises plasticity with linear hardening, International Journal for Numerical
Methods in Engineering 56 (2003), 1375-1396.
4 Caddemi S., Computational aspects of the integration of the von Mises lin-
ear hardening constitutive laws, International Journal of Plasticity 10 (1994),
935-936.
5 Chan A.H.C., Exact stress integration for von Mises elasto-plastic model
with constant hardening modulus, International Journal for Numerical and
Analytical Methods in Geomechanics 20 (1996), 605-613.
6 Cocchetti G., Perego U., A rigorous bound on error in backward-dierence
elastoplastic time-integration, Computer Methods in Applied Mechanics and
Enggineering 192 (2003), 4909-4927.
7 Dominici D., Nested derivatives: a simple method for computing series
expansions of inverse functions, International Journal of Mathematics and
Mathematical Sciences 2003 (2003), 3699-3715.
8 Gratacos P., Montmitonnet P., Chenot J.L., An integration scheme for
PrandtlReuss elastoplastic constitutive equations, International Journal for
Numerical Methods in Engineering 33 (1992), 943-961.
9 Hong H.K., Liu C.S., Prandtl-Reuss elastoplasticity: on-o switch and
superposition formulae, International Journal of Solids and Structures 34
(1997), 4281-4304.
10 Kossa A., Szab L., Exact intergration of the von Mises elastoplastic-
ity model with combined linear isotropic-kinematic hardening, International
Journal of Plasticity (2008), available at http://dx.doi.org/10.1016/
j.ijplas.2008.08.003.
11 Krieg R.D., Krieg D.B., Accuracies of numerical solution methods for
the elastic-perfectly plastic model, Journal Pressure Vessel Technology 99
(1977), 510-515.
12 Liu C.S., A consistent numerical scheme for Mises mixed hardening consti-
tutive equations, International Journal of Plasticity 20 (2004), 663-704.
13 Nemat-Nasser S, Plasticity. A Treatise on Finite Deformation of Heteroge-
neous Inelastic Materials, Dover Publications, Cambridge Univ. Press, 2004.
14 Reuss E., Bercksichtigung der elastischen Formnderung in der Plastiz-
ittstheorie, Zeits. angew. Math. Mech. (ZAMM) 10 (1930), 266-274.
15 Ristinmaa M., Tangent modulus in Finite Element Calculations for Non-
Linear Materials, LUTFD2/(TFHT-3033), Div. of Solid Mechanics, Lund
University (1990).
16 Ristinmaa M., Tryding J., Exact integration of constitutive equations in
elasto-plasticity, International Journal for Numerical Methods in Engineering
36 (1993), 2525-2544.
17 Romashchenko V.A., Lepikhin P.P., Ivashchenko K.B., Exact solution of
problems of ow theory with isotropic-kinematic hardening. Part 1. Setting
the loading trajectory in the space of stress, Strength of Materials 31 (1999),
582-591.
18 Simo J.C., Hughes T.J.R., Computational Inelasticity, Springer, Berlin,
1998.
19 Spanier J., Oldham K.N., An Atlas of Functions, Springer, Berlin, 1987.
20 Szab L., A semi-analytical integration method for J2 ow theory of plastic-
ity with linear isotropic hardening, Computer Methods in Applied Mechanics
and Engineering Submitted (2008).
21 Szab L., Kovcs ., Numerical implemetation of Pragers kinematic hard-
ening model in exactly integrated form for elastic-plastic analysis, Comput-
ers and Structures 26 (1987), 815-822.
22 Wei Z., Peric D., Owen D.R.J., Consistent linearization for the exact stress
update of PrandtlReuss non-hardening elastoplastic models, International
Journal for Numerical Methods in Engineering 39 (1996), 1219-1235.
23 Xucheng W., Liangming C., Exact integration of constitutive equations of
kinematic hardening material and its extended applications, SMIRT-8, Brus-
sels, Proc. Paper L2/3 (1985), 65-70.
24 Yoder P.J., Whirley R.G., On the numerical implementation of elastoplastic
models, Journal Applied Mechanics 51 (1984), 283-288.
Integration method for constitutive equation of von Mises elastoplasticity with linear hardening 75 2007 51 2